Mutual Investments Funds in Shares in Brazil : Incentives , Management and Convergence

This article analyzes patterns of convergence in series of return of investment funds in shares in Brazil, from the use of the methodology of Phillips and Sul (2007). Unlike the theory of portfolios and in disagreement with the characteristics of this market regulation, transparency, efficiency and informational transaction costs -, it shows a heterogeneous behavior, from the formation of four clubs with specific transition and composition dynamics. The private nature of the management legal entity, the incentives associated with the recovery of low administrative fees, in addition to rates of performance and the emphasis on gains offset by risks, differentiate funds that are losers or winners.


INTRODUCTION
he relevance of the financial market in a society can be associated with the use that economic agents make of the products and services available, with a view to a more intertemporally efficient allocation of resources and over the possible uncertain states of nature, and, for this reason, many existing instruments, with different characteristics and idiosyncrasies.
In this context, the gradual growth and sustainable versatile and diversified "industry of mutual investment funds" is observed not only in more developed countries, but rather as a world trend.In keeping with the Latin America, for example, the total volume of assets under management of investment funds, at the end of 2010, was of 1.13 trillion dollars, an amount exceeding 30% of Gross Domestic Product (GDP) of these countries, according to data from reports of the International Monetary Fund, the Investment Company Institute (ICI) and the Federación Iberoamericana de Fondos de Inversión (FIAFIN).This percentage aggregate is captained, mainly, by the Brazilian economy, with ratio close to 50 %, it is possible to observe that this relationship in developed economies is still well above it.In France and in the United States, the volume managed by funds reaches 65% and 80% of their respective GDPs.In addition to moving volumes representative in these developing countries, this industry shows annual growth of net equity of 11 %, during the years 2007 to 20101 .This evolution consists in a reality -unlike evidenced in the case of other financial assets less anchored in foundation -with theoretical counterpart, just to do this, observe one of the consequences of the Ownership of the Separation addressed in classical work and pioneering developed by Markowitz (1952).
In Brazil, the closing of 2010, supplied by the Brazilian Association of Entities of Financial and Capital Market (Associação Brasileira das Entidades dos Mercado Financeiro e de Capitais -ANBIMA), indicate that the country is the sixth largest market in the world, with more than 10,000 products and approximately 11 million investors, being the net worth over $1 trillion, with annual growth rates of the order of 20 %.BBR, Braz. Bus. Rev. (Engl. ed., Online), Vitória, v. 12, n. 2, Art. 6, p. 110 -142, mar.-apr. 2015 www.bbronline.com.br Despite the numerical evidence about the relevance of this segment, there is still a gap in the evolution of specific theoretical and empirical finance literature designed to measure the performance, to modeling, to pricing, to forecast returns and prices or even the analysis of contracts with punishments and incentives typical of the management of mutual investment funds.
Among the few articles found in the international literature, it is highlighted the work of Fama and French (2010), according to which, in comparing portfolios formed by funds and portfolios of traditional market by means of simulations via bootstrap, few funds would be able to generate returns compatible with industry benchmarks to point to compensate for the costs of active management characteristic of the funds.
For the Brazilian case, an interesting and comprehensive survey of descriptive was written by Varga and Wengert (2010), with emphasis on the evolution of the market for investment funds, its regulatory scenario, its organization and the types of investors.In keeping with the derivation of pricing models and forecasting, possibly one of the first jobs was that of Matos and Rocha (2009).According to the authors, the models of linear factors capturing better than Capital Asset Pricing Model the risks associated with the investment funds with higher net worth and with greater gap of performance in relation to the Ibovespa.As a result of this aspect of modeling specific pricing of funds include Matos, Zech and Linhares (2012), and Matos and Silva (2011).
Still in the segment of Brazilian funds of investment in stock, Matos and Castro (2012) explore the construction of dynamic portfolios composed only by funds, according to various risk-return performance metrics , while Matos, Padilha and Benegas (2013) examine the question under aspects related to frontier efficiency associated with accounting headings of active and passive funds in years of economic boom and crisis.
Adding to this literature on funds in stock\shares, to observe the evolution of actual return accumulated panels of these assets for long periods of time, the researcher might be interested in examining the evidence on the convergence or non-convergence of such time series, in addition to wanting to analyze which factors could be useful in the understanding of such behavior.
In this sense, this Article consists of a work of an exploratory, once it is not necessarily based on a theoretical framework that constitutes the need for a behavior convergent or not.
The article deals with empirical issues associated with the effects caused by incentives, by BBR, Braz.Bus.Rev. (Engl.ed., Online), Vitória, v. 12, n. 2, Art. 6, p. 110 -142, mar.-apr. 2015 www.bbronline.com.brnature and experience in the management of the fund, by size, by risk-return performance, by the metrics of risk and by other administrative variables capable of characterizing robust evidence that, even in a period of high efficiency informational and financial dynamism, this competitive industry of mutual investment funds in shares in Brazil presents paradoxical high levels of divergence.Following Antzoulatos et al. ( 2008), Higson, Holly and Petrella (2009), Caporale, Erdogan and Kuzin (2009) and Matos, Penna and Landim (2013), it is here used of the semiparametric technique proposed by Phillips and South (2007) -robust heterogeneity present in the funds over time market, by relying on the formulation of a model of nonlinear factors idiosyncratic element composed of a time-variant factor and a common stochastic trend -applied to a monthly panel containing 68 funds in shares during the period from January 1998 to June 2007.
The lack of a common trend is reinforced by the evidence of the formation of four clubs of convergence, whose transition dynamics and composition are quite specific.The results suggest that the compositions of the distinct groups of funds follow standards so that better positioned funds are mainly associated with the following variables: greater financial incentives generated to the management of the fund and from contracts that foresee the rates of outperformance vis-à-vis benchmarks on stock market or metrics of inflation, lower values of administrative fees, shorter maturity or time of operation, significant presence of private financial institutions in the management, emphasis in the management of portfolios in metrics of risk-return performance.The effects caused by the size of the fund or the management on the basis of other moments focusing the distribution of the returns do not seem to be very useful in the characterization of the groups.
The article is structured as follows: section 2 deals with a concise history and relevance of the market for investment funds in shares in Brazil; section 3 is a review of the literature in terms of framework and fund market; section 4 describes, in detail, the methodology for identifying the clubs of convergence; section 5 presents the empirical exercise, in addition to the discussion of the results.The final considerations are made in the sixth section.Before ascension of investment funds, incentives were granted, such as the regulatory innovation of 1967.Then, the Fundos Fiscais de Investimento em Renda Variável were created, popularly known as Fundos 157.Still according to Oliveira Filho ( 2008), the decree granting tax incentives for companies to launch shares on a stock market and for which the individuals invest in these roles, having represented one of the major driving forces in the capital market and the mutual fund industry in Brazil.

HISTORY
However, the macroeconomic imbalances, as the speculative crisis of 1971 in the stock markets and the oil shock in 1974, were the decisive factors for the decline of the fund industry, which trajectory was interrupted by the appearance of fixed income funds in the 1980s, coinciding with the Banco Central's role in combating inflation.In this context, the 1990s are a milestone toward the expansion of investment funds.First, the ANBID consolidates itself as the main provider of information in the capital market, creating the first counting profitability rate of funds system, with the adoption of single criterion for the market as a whole.Secondly, there is the gradual liberalisation and opening up of the financial Another important point is the internationalisation of the financial system, with the presence of foreign banks in Brazil, change that brought the segmentation of activities in the administration of funds of the countries of origin, considered more appropriate financial systems with universal banks.
Even in the 1990s, with the objective of encouraging the financial investment, was created the Financial Investment Funds (Fundo de Investimento Financeiro -FIF) and a fund of funds, known as Fundo de Aplicação in Cotas de FIF, responsible for absorbing all the current fixed income funds.
From the year 2000, the mutual fund industry has gone through a number of liberalizing reforms, whose aim was to increase the freedom of management of funds and stimulate the differentiation of the segment.During this period, the Comissão de Valores Mobiliários has taken on greater responsibility for the monitoring and supervision of funds, at the same time that the Associação Nacional de Bancos de Investimento -ANBID has developed the Code of Self-Regulation.Already in 2004, the CVM emits its Instruction no.409, currently in force, which constitutes the main legal device to regulate the investment funds in Brazil.
Features on the constitution, the administration, the functioning and the dissemination of information, in order to enhance the efficiency and improve the standards of monitoring and supervision.

RELEVANCE OF THE MARKET OF FUNDS
Possibly, a way to justify the study of a segment is to analyze its evolution over time and the order of magnitude of the amounts involved.Thus, by observing the data of the market, it is possible to realize that the amount of investment funds operating regularly increased from approximately 1750 to almost 4800, in an interval of 11 years, an average annual growth rate of 9.8% per year, which was accompanied by an average annual increase of 23.8% in net worth household, passing this industry to move almost R$ 1.5 trillion at the end of 2009, while this volume, at the end of 1998, was only R$ 150 billion.
So, there is a market with growth rates higher than those observed in the real economy, in addition to amounts comparable with the own National Gross Domestic Product, which

STUDIES ON FINANCIAL INTEGRATION
Assuming that there is an almost perfect correlation between the international markets shares in long time horizons and that these markets share a common trend, was developed a literature of financial integration, according to which, the returns of shares adjusted by risk recorded in a common currency would be the same in all countries.
This approach had its beginning with the analyzes Von Furstenberg and Jeon (1989), involving the return of daily actions from 1986 to 1988 for the USA, Japan, Great Britain and Germany.The principal evidence were about the strong correlation and trend of joint development, reinforcing the concept of financial integration between the international markets.
The continuity was given by Littell (1997), Securato (1997) and Lombard, Roulet and Solnik (1999), who found the same evidence, suggesting that the level of development of the countries (or another similarity between them), can keep some relationship with the process of integration.Making use of other techniques, but with the same purpose, also stand out Kasa (1992), Corhay et al. (1993) and Rangvid (2001).All these analyzes assume that all markets are perfectly integrated, or perfectly segmented, or even partially integrated, but with a degree of integration relatively constant.Common to all these studies, is the assumption from a theoretical framework about the financial integration between financial markets situated in several economies.While there is no direct applicable and fundamental counterpart to an internal market composed of investment funds in the same category, analyzing the main informational aspects, accounting, financial and regulatory changes comes to be a reasonable question about the level of integration and competitiveness among the components of this industry.

THEORETICAL FRAMEWORK
In the literature of Theory of Assets Pricing, the biggest challenge is possibly associated with the ability to develop preferably microfoundations frameworks such that are accommodated the main empirical evidence in more diverse financial markets, in temporal dimensions or in cross-section.Surveys such as Cochrane (2001Cochrane ( , 2006) ) are excellent sources on pricing models of individual actions, and can demonstrate that, in spite of very legitimate, this excessive concern can be put in the background, other markets, mainly the investment funds, recognized as relevant and with one of the largest potential for growth.
Faced with this evidence on a sort of absence of a specific literature that stick to mutual funds for investment in shares, or even investment funds in other categories, it is essential to ask what could be developed in terms of theoretical modeling to price or to predict the returns of these dynamic portfolios, on the history of the industry, or even management techniques.This literature still little explored, in addition to the contribution of Fame and French (2010) comparing portfolios of funds and portfolios of traditional market, it is important to highlight Haslem's (2010).The author covers from the historical aspect, passing through the services offered and the regulations on the industry of American funds, until the analysis of tools for managing the funds.Interesting working papers in this area, but more focused on analysis of composition of portfolio and performance, and not necessarily modeling pricing, are Adcock et al. (2009), Bessler et al. (2010) and Ferreira et al. (2010).
Still on the behavior of investors, it is worth quoting Rubbany et al. ( 2010) and Brookfiled and Cortez (2010).

Modeling of Pricing of Investment Funds in Brazil
Pointing to the pricing and forecast of returns, one of the first papers consists of Matos and Rocha (2009), which highlights that the models of linear factors similarly to Fama andFrench (1992) and(1993) capture better than the CAPM -originally developed by Sharpe (1964), Lintner (1965), Mossin (1966) -the risks associated with the investment funds with higher net worth (PL) and with larger gap of performance in relation to the Ibovespa, although it does not appear to be that sufficient improvement.This motivation was followed by Matos, Linhares and Zech (2012), according to which there is evidence of the existence of dynamic behavior distinct for the return of investment funds, over time, depending on the non-linearity characteristic when Threshold CAPM (TCAPM) is used, in the same way, for example, the  Tong (1978) and by Tong and Lim (1980).
An interesting and successful extension was developed in Matos, Correa and Silva (2012), which tested in exercises of pricing and forecasting in-sample, a framework model of linear factors.Following the methodology developed in Fama andFrench (1992, 1993)

Management and Performance of Investment Funds in Brazil
Specifically on the management of portfolios, an interesting study applied to Brazil consists in Varga (2001), which implements various statistical measures of performance assessment to the ten largest stock funds offered in the Brazilian market, showing the different classifications of funds according to each of these indicators.Already Varga and Wengert (2003) identify business strategies common to investment funds offered in the Brazilian market, while Nephew (2003) studied the effectiveness of active management of investment funds in shares to obtain, in a consistent and lasting manner, superior returns than Ibovespa, noting the risk that these funds incurred and the ability of its managers to buy stock understated and diversify their portfolio.According to this work, 70,46% of these funds have not been able, by means of a pro-active management, overcome the Ibovespa. 4onteiro ( 2006) evaluates 112 Brazilian funds through measures of profitability, selection of assets, market timing and measures that combine the last two, with daily data in windows of three months and a year, for the period of January 1998 to December 2005, using the methodology of Bollen and Busse (2005).The evidence suggests the existence of persistence of performance by selectivity and market timing for up to six months for a small

CONTRIBUTION
In spite of the unquestionable relevance to discuss the modeling of pricing and forecast of future returns, or even the question of the management of portfolios managed by investment funds, investors and financiers are before an industry with not yet explored idiosyncrasies, in which, by regulation, is a need for at least 67% of the portfolio be composed of shares traded on the São Paulo Stock Exchange (Bovespa), which limits the possibilities in this dynamic and competitive market, and which financial information, administrative and accounting are all disclosed on a monthly basis on CVM and, consequently, were made available to the public.
Thus, there is a market with a high level of regulation and transparency, characterized by low costs associated with the transaction of purchase of quotas, with change itself on the fund to invest and which divergence in terms of return between distinct funds would not be expected by the Property of Separation developed in Markowitz (1952) -the differences between the optimal risky portfolios in each fund could be associated with the restrictions imposed by expertise of management or technical capability or financial volume available by the fund.From the perspective of the investor, only an inertia or the low level of financial knowledge could be partially explaining the imbalance evident when the sharp divergence between returns of different funds.In this context in that there are aspects that suggest that there is some degree of integration in the market of investment funds in shares, from the development of methodologies for the analysis of convergence, it is important to address the issue of the existence or not of a common trend of cumulative return.Was there in fact a common trend?If not, what standards could be associated with the formation of clubs This level of accumulated return is vastly superior to its gain of the Ibovespa, at 188 %.
There is an unusual amount of funds with underperfomance in relation to the same benchmark of market, about 27 of the 68 funds, the worst accumulated return being the Alfamais Fund (alpha), with less than 35% of accumulated gain.An alternative in terms of statistical framework would be the use of the analysis Q, building typology, analysis of classification, or even cluster analysis, a technique that aims to (internal) homogeneity of objects within the same group and the (external) heterogeneity between the groups.Unfortunately, in spite of the versatility of the technique, this is very sensitive to the sample used and generally characterized as descriptive, not been a greater theoretical basis and not been possible to ensure the uniqueness of the results, or even allow outline statistical inferences.In this sense, this article analyzes the patterns of convergence in a series of actual accumulated return for a panel with 68 investment funds in shares in Brazil, identifying a set of financial variables, accounting and administrative provisions which seem to be able to characterize the possible clubs of convergence.
In quite different contexts, but aligned conceptual and methodological approach to this article in the sense of the need to examine the convergence or divergence of temporal series, confirming or rejecting theories about the behavior of long-term series, is worth mentioning Penna and Linhares (2009) on empirical exercises of trajectories of GDP, Matos, Penna and Landim (2011), about an application of Phillips and Sul (2007) for stock exchanges, or even Matos, Penna and Vasconcelos (2013), whose object of study was the credit market.Here, the use of this non-linear framework aims to model the heterogeneous behavior of idiosyncratic elements associated with the funding of investments in shares in Brazil.As the contribution of this study on the discussion of Brazilian industry is isolated and restricted to the temporal dimension, a natural extension and interesting is to test the hypothesis from Hansen (2000) the convergence or divergence in cross-section size.See Penna and Linhares (2013) for applications of this nature., through the following relationship: The empirical strategy of Phillips and Sul (2007) consists in modeling the data panel in such a way that the common components and idiosyncratic could be distinguished, that is, In that t  it is a component that determines the trajectory of long-term, i.e., a common history of growth of accumulated gain by funds, and t i b , it is an idiosyncratic element that varies in time, capable of measuring the individual effects of transition. 5 In these terms, it will be possible to test the convergence of long-term ( ) where the heterogeneity not observable to dissipate, that is to say, where The inferences about the behavior of t i b , are not possible without the imposition of any restriction in its dynamics, because the number of unknown parameters in t i b , is equal to the number of observations.Thus, an alternative to model the elements of transition can be derived from the construction of a coefficient of transition on, t i h , , defined as: Where t i x , ˆ represents the accumulated gain of the fund without the component of economic cycles. 6Thus, the curves drawn by t i h , define a trajectory of transition on and, at the 5 t i b , Can be idealized as the trajectory of the individual transition of i , given your travel around the common trajectory, t  , been necessary to emphasize that, although this heterogeneity among the funds, the financial market still holds common characteristics that compound them; such common characteristics can be influenced in any permanent contagion effect or cultural, technological (such as argue Phillips and Sul in macroeconomic models), institutional, socioeconomic, governmental factors and other not observable factors, hence the assumption of common component. 6In practice, the variable used can be described as (4) Thus, on the basis of this modeling, Phillips and Sul (2007)  The authors assume that the coefficients of transition are estocastic linear trends and allow heterogeneity between the trajectories over time of gain in each fund.
To model these coefficients, it is proposed the follows semiparametric form: is a slowly varying function increasing and diverging from the infinity ,  governs the rate of decrease in the variation of the cross-sectional units over time and , then this formulation ensures that It is possible to establish a test of the null hypothesis of convergence against a nonconvergence alternative hypotheses.This test is based on the following assumptions: Null Hypothesis: Alternative Hypotheses: Hodrick-Prescott (WHP) filter.This approach does not require any specification a priori to t  and it is very convenient, because it requires a single smooth parameter as input.BBR, Braz.Bus.Rev. (Engl.ed., Online), Vitória, v. 12, n. 2, Art. 6, p. 110 -142, mar.-apr. 2015 www.bbronline.com.brThis approach also allows testing the formation of clubs of convergence.For example, there are two clubs , The alternative hypothesis can be described as follows: Under the null hypothesis, it can be inferred about the significance of the coefficients of ( 7) on the basis of a test t that is unilateral, robust to autocorrelation and heteroscedasticity.For a level of 5 %, for example, the null hypothesis of convergence must be rejected if 1  t < -1.65.For that initial observations do not exert strong influence on the results, Phillips and Sul suggest that the regression ( 7) is estimated after discarding a sample fraction.Based on the Monte Carlo simulations, these authors suggest that, in order to achieve ideal properties in terms of size and power, the relation ( 7) is that regressed after cut, approximately one-third of initial observations.Finally, the rejection of the null hypothesis of convergence for the entire panel can indicate the existence of separate points of balance or multiple stationary states.When this occurs, you can take the divergence of some states or the formation of clubs.In this context, an algorithm that applies sequentially the test t log enables the identification of clubs of convergence without recourse to usual observable characteristics that prejudge the grouping this club.The algorithm is in the appendix.

DATABASE
Before trade-off usual during implementation of empirical exercises, as regards the data base in its temporal dimensions (T) and the cross-section (N), we decided to use the complete time series that compose the available database on the web site www.fortuna.com.br.On this basis, there is information of prices, net worth and cadastral/administrative data of stock funds in Brazil.Thus, we chose to use the funds classified by ANBIMA as Stock Investments, type Ibovespa Ativo.Been limited to funds with complete series between January 1998 and June 2007, 114 monthly observations, 68 funds were selected.A lesser quantity of months would not be advisable due to the asymptotic aspect of the methodology, while the adoption of a larger quantity of months could considerably reduce the cross-section of funds used.It is important to note that the framework used is not robust to the presence of structural breaks, so that the strong declines in the evolution of the temporal series analyzed -evidenced in part in the second half of 2007 and more strongly from 2008, due to the financial crisis -limited the use of a better dashboard, which could comprehend until the month of December 2009, although it had not been reduced the amount of funds.
In addition, because they are funds of the Ibovespa active type, using it as a benchmark of the traditional Stock Exchange Index values of São Paulo (Ibovespa), for purposes of direct comparison of metrics of expected gain, risk and performance.
In order to make both the series of monthly returns and accumulated return real, it was made using the Índice Nacional de Preços ao Consumidor Amplo (IPCA), which source was the institution website itself, the Brazilian Institute of Geography and Statistics (IBGE).

DESCRIPTIVE STATISTICS OF FUNDS FOR INVESTMENTS IN SHARES
Possibly, a more detailed manner to analyze the divergence -in which the fund Prosper Adinvest (pros) reaches a real return net of accumulated 1,580 %, while the Ibovespa reaches the level of 188% and the Alfamais (alpha) only the order of magnitude of 35 %, during the period from January 1998 to June 2007 -is to observe not only the evolution of earnings accumulated, but the moments of the distribution of returns, as well as the associated metrics performance and size, these statistics are reported in Table 2.
In this table, are defined the codes associated with the respective funds, bearing in mind the aim to facilitate the exposure of the them in other tables used in the course of this work.
Over the same period analyzed in this article, while assets taken as a benchmark of the money market, more specifically of fixed income, such as investments in savings and shortterm title of the Brazilian government yielded approximately 4% and 13% a year, and had annual volatility of the order of magnitude of 7% and 9 %, respectively, are to be found, of course, values much higher than for the large majority of funds selected.In the capital market, the traditional Ibovespa index yielded, on average, more than 18% a year, but having as counterpart a volatility higher than 70 %.
Except the pros and small funds, which average real monthly earnings were above 2.5 %, there are 12 funds with gains ranging between 1.5% and 2% monthly and 21 others with As regards the metrics of risk, there is no clear trend in standard deviation, in the market beta, not even in the downside risk, the latter obtained from the Ibovespa Index as a benchmark.In other words, little can be inferred about the behavior of total risk and systemic between funds arranged in descending order in accumulated gain.Nevertheless, an interesting evidence consists in the clear pattern for the performance metrics used.The Sharpe Indexthe most common of all and initially called reward-to-propose ratio, which origin is associated with Sharpe (1964) -consists of the ratio between the risk premium paid for the share in question and its volatility, if this can be adequately approximated by the standard deviation. 7 In the case of the Treynor Index me due to the contribution of Jack L. Treynor (1965), this metric measures the compensation of additional gain on the risk-free asset for systemic risk unit (instead of total risk also incorporates the idiosyncratic risk) of incurred market being captured by of the market, obtained when using the CAPM regression.
Observing these two metrics, there is a tendency toward a reduction of performance for funds with less accumulated gain, evidence that may indicate that the first two moments of the distribution relevant to the analysis made by the investor, justifying the fact that funds taken as winners were more focused on such metrics.In more precise analyzes of portfolio management, it is usually preferable the use of excess return as a compensation of pricing model based risk, such as the CAPM or linear factors, or even the use of the Index of Sortino, proposed by Brian Rom in 1986, that provides a value for the compensation of additional gain regarding a minimally attractive benchmark per risk asymmetric unit, which penalizes only deviations below the mean or the defined reference, unlike the standard deviation which penalizes deviations from good and bad surprises.
Thus, as regards the alpha of Jensen, once again it shows a behavior to increase of this metric of performance relative to the increase of accumulated gain, the same was not observed for the index of Sortino greater than the Ibovespa Index in the same period, result accompanied by a systemic risk slightly less than 1.0, according to the beta of the market of this industry, which median in terms of PL suggests a level higher than the R$ 20 million.

IDENTIFICATION OF CLUBS OF CONVERGENCE AND THE PROVISION OF TRANSITION
The methodological procedures described in section 4 were applied to a panel containing the monthly cumulative return of 68 funds for investments in shares, during the period of January 1998 to June 2007, including even at the Bovespa Index, in order to be able to identify in which club this index would be in, in addition to the evidence that funds converge to the benchmark, a sign of passive strategy of this fund.Initially, we tested the overall convergence of accumulated return of funds by means of the equation ( 7). 8 The results indicate an estimation of 1  equal to -1.2163, with its statistical t of - 22.0158, value of less than -1.65, rejecting the null hypothesis of global convergence to a common level.The first evidence is, therefore, to confirm that there is no convergence of accumulated gain in market of investment funds in shares in Brazil.
Been discarded the hypothesis of absolute convergence, it is possible that there will be the return of a group of funds is converging.Thus, has continued the procedure previously described for the identification of possible clubs of convergence.Due to the small size of the sample, we sought a greater parsimony in the determination of the clubs, by setting 0 *  c . A summary of estimates obtained for the clubs of convergence identified is shown in Table 3 Vitória, v. 12, n. 2, Art. 6, p. 110 -142, mar.-apr. 2015 www.bbronline.com.brsequence of analysis suggested the end formation of 3 other clubs of convergence, while the isolated alpha fund, not have converged to none of them.Longer-term trends of 4 clubs are presented in Figure 1, with emphasis on the detachment for the first club.9 3 that the second club is composed of 21 funds, there is no public institutions as managers, but the presence of two of the largest and most traditional banks in the country: Bradesco and Itaú.In Figure 5, in which they were portrayed the transition dynamics of the funds that make up the second club, calls attention to the presence of the fund voto1, which presented the lowest medium real liquid return of the club, in addition to the smaller indices of The third club consists of 20 funds, which converge together as the Ibovespa, being timid in presence of public institutions in the management of component funds, in addition to the striking presence of Banco Itaú, with 24% of the funds.In Figure 6, the Ibovespa appears as a component of the third club that has a dynamic that presents higher end value, and there is still an apparent detachment by lower values of convergence because of the fund unib4.
The club room is composed of remaining funds for smaller performances accumulated for the considered period, being even more significant participation of public banks fund management components of this last club, of approximately 15 %.

DISCUSSION OF RESULTS GENERAL
An analysis that seeks to detect common patterns of behavior in long and volatile time series of returns in samples with a high amount of financial assets, from the descriptive statistics in Table 2, for example, it is often inaccurate and inconclusive, if we do not make use of a statistical technique to reduce the sample of N assets for K groups of assets.
Thus, on the basis of the grouping obtained from the technique of Phillips and Sul (2007), already described in section 4, the composition of which is reported in A specific analysis of the third club suggests that there are 1/3 of the funds assuming market efficiency and adopting a passive strategy following the benchmark of the market and only featuring gentle losses in relation to this, because of failed active strategies.A strong signal, if it reflects the reality, of irrationality on the part of economic agents that do not punish the loser funds, changing positions and migrating to the systematically winner funds.
On the basis of Table 4, as would be expected, there is a clear distinction between the actual monthly average returns of clubs, respectively, 2.38 %, 1.76 %, 1.36% and 1.05 %, been the distance between the first club and others the greater distance, which corroborates the offset by higher values in the dynamics.On the net worth of each club, evidence-if an outlier behavior on the part of the first club, which has average values of the order of 3 to 4 times the PL of other clubs, even possessing funds with medium PL below R$ 10 million.
According to the metrics of risk used, except when the use of downside risk, the clubs seem to have a pattern of increasing risk to the extent that have less accumulated gain.10Paradoxically, the funds with the greatest gains have even lower levels of total risk and systematic.Already abide by the downside risk, the second club would be the less volatile, while the first club would be the highest risk.
Analyzing the third and fourth moments of the identified clubs, as well as in disaggregated analysis, the coefficient of asymmetry suggests clubs invariably with asymmetry to the left, with lower intensity in the first club, and they all moderately dense tails.
Due to higher expected returns and lower levels of risk, the component funds of clubs considered as winners not only hit the market, as the other funds, but exhibit the highest performance metrics, been the order of magnitude of the indexes of Sharpe and Treynor of the first club of two to 3 times the level of the third and fourth clubs.Considering the Index of Sortino, the best performance would be the funds from the second club.
All these evidences are financial corroborated when in the even more aggregate analysis from equal-weighted portfolios formed from the funds of each club, whose statistics are reported in Table 5 and whose evolution is in Figure 2. It is worth emphasizing the detachment even more evident in portfolio #1 composed of the 14 funds of the first club, which, in addition to having better performance up to June of 2007, is the one that presents drawdown which order of magnitude is less than that of the Ibovespa Index of other portfolios, having been the only portfolio capable of recovering the loss with the financial crisis, until December of 2009, a strong evidence in favor of the expertise of active management of private institutions that adopt incentive mechanisms and punishment.
In relation to administrative variables, it is noted that the administration rate of 68 funds in this study had a great breadth, ranging from 0.40% to 7.5% and that, on average, administration rates are larger the smaller the profitability of clubs.The rate of total performance, only 12 of the 68 funds charge this type of fee, which generally consists of percentage ranging from 15% to 25% of the that exceed the Ibovespa or the General Price Index of the Market (Índice Geral de Preços do Mercado -IGP-M).An important observation is that all the funds that adopt performance rate -this type of mechanism with variable incentive to the manager who consistently beats the market or the established targetare invariably in winner clubs, an evidence of that, having no punishment on the part of investors in terms of the redemption of the amount applied on the fund by reason of bad results, the design of mechanism should provide for a contract with variable gain to the manager.
According to these data, the selected funds started its activities, on average, 17 years ago, and there is a curious inverse relation of profitability of clubs with the maturity of the same.Finally, there is only the presence of private financial institutions as managers of the funds that make up the first two clubs, with emphasis on Banco Safra, with 15% of the funds from the first club, the Banco Itaú and Bradesco both representing 14% in the formation of the 2nd club.

CONCLUSION
In the literature of investment funds in shares, it is discussed the ability of managers to opt for active strategies able to obtain accumulated gains greater than the market.There is no consensus in this still little developed literature, what would be expected in a market with a high level of regulation and transparency, characterized by low costs associated with the transaction of purchase of quotas, as the change of funds itself?
, they built factors, which consist of zero cost equal weighted portfolios composed only by funds, capable of capturing the size effects and performance of these assets, being the same as those used in various applications in an extended version of the CAPM, for a panel composed of the 75 investment funds in shares in Brazil, from January 1998 to December 2008.The size effects and performance evidenced by the inadequacy of the CAPM model in funds with higher net worth and very high or low performances, seem to be very well accommodated when the incorporation of factors, which showed significant, isolated and jointly, in almost 50% of the 75 funds analyzed.The main evidence obtained from regressions individual temporal are supported when the test in panel with random effects, in that both effects are indispensable in the explanation of the returns of the funds for investment in shares in Brazil.
has the two conditions for convergence of the model.

Figure 1 -
Figure 1 -Long-term Trend of the four clubs of convergence identified Note in Table3that the second club is composed of 21 funds, there is no public institutions as managers, but the presence of two of the largest and most traditional banks in the country: Bradesco and Itaú.In Figure5, in which they were portrayed the transition dynamics of the funds that make up the second club, calls attention to the presence of the fund voto1, which presented the lowest medium real liquid return of the club, in addition to the smaller indices ofSharpe and Treynor.
Thus, the first Brazilian mutual fund, the CRESCINCO, dated 1957 -beginning of Juscelino Kubitschek's government and the developmentalist era -and had as objective channeling the funding to the nascent Brazilian base industry, according to Oliveira Filho(2008).Varga and Wengert (2010)indicate that, in 1959, the current Ministry of Finance issued the first official document to deal with mutual funds.A great advance was made in the market of investment funds in Brazil since the 1960s, with the reform of the national financial system, especially in the period from 1964 to 1966. 2 The establishment and development of the National Financial System occurred in fact from the year 1964, with the creation of the National Monetary Council (Conselho Monetário Nacional -CMN), the Central Bank of Brazil (Banco Central do Brasil - Throughout history, the market for investment funds in Brazil and its role in economic development has been gaining importance and interest.The investment funds, according to Alves Júnior (2003), were direct fruit from the conduct of economic policy, and its initial development associated with the regulatory initiative in the late 1950s was very BBR, Braz.Bus.Rev. (Engl.ed., Online), Vitória, v. 12, n. 2, Art.6, p. 110 -142, mar.-apr.2015 www.bbronline.com.brincipient.BCB) and in 1967, when it began in Rio de Janeiro the National Association of Investment Banks (Associação Nacional dos Bancos de Investimento -ANBID).Already in 1976, it was established that the Comissão de Valores Mobiliários (CVM), these were some of the milestones of early standardization of specific operation of investment funds in Brazil.

Table 1 -Participation of Categories on the Market of Investment Funds in Brazil (%) A
Matos and Castro (2012)ults for the 10 per cent of best performance of the sample.Duvernoy (2009), using stochastic factors of allowable non-parametric discounts, does a study on the performance of investment funds in the Brazilian industry.Under the hypotheses of incomplete markets and non-arbitration, are certain limits to the performance of funds when based on admissible measures.In addition, the author presents a variation of this method in which is excluded the possibility of very high Indices of Sharpe persisted in a market in balance.In a monthly sample of 33 multimarket funds, for a period of approximately eight years, the results show that the performance of funds can vary greatly depending on the measure.Even for the segment of Brazilian funds investment stock,Matos and Castro (2012)explore the construction of dynamic portfolios composed only by funds, according to various risk-return performance metrics.
BBR, Braz.Bus.Rev.(Engl.ed., Online),Vitória, v. 12, n. 2, Art.6, p. 110 -142, mar.-apr.2015www.bbronline.com.brnumber of funds Would there be a relationship with financial and administrative aspects of management of funds?The simple observation of the evolution of the returns accrued makes it possible to highlight the signs of imbalance in this market during January of 1998 to June of 2007, in which, in addition to the pronounced divergence, there is a scenario with very high outperformance in relation to the Ibovespa, with the Prosper Adinvest Investment Fund (pros), generating a real aggregate gain net of 1580 %.
, measure the extent to which the gain of the fund i moves in relation to the trajectory of common growth.t  .In this way,  represents an effect of business cycle.The removal cycles of the cycle component can be performed through the use of a Whittaker-BBR, Braz.Bus.Rev.(Engl.ed., Online),Vitória, v. 12, n. 2, Art.6, p. 110 -142, mar.-apr.2015www.bbronline.com.brsame time developed a convergence analysis based on what they called test t log .

Table 2 -Main Descriptive Statistics of Investment Funds in Shares In Brazil A, B, C (Continued)
a Painel contendo séries temporais mensais de retornos reais líquidos de 68 fundos de investimento em ações(1998.1 a 2007.06,114observações)cFonte dos dados: Site www.fortuna.com.brbDisposição dos fundos: ordem decrescente por retorno real líquido acumulado BBR, Braz.Bus.Rev.(Engl.ed., Online),Vitória, v. 12, n. 2, Art.6, p. 110 -142, mar.-apr.2015  www.bbronline.com.br . In relation to the third and fourth moments, the great majority of PLs in the order of R$ 5 million to R$ 7 million with performance very close to funds with PL of approximately R$275 million.In order to summarize the information in the sample, the observation of median values reported at the end of Table2, robust to the presence of outliers, allows us to infer an accumulated gain, of the order of almost 250% 7 Only in 1994 this metric was entitled with the name of William Forsyth Sharpe.BBR, Braz.Bus.Rev.(Engl.ed., Online),Vitória, v. 12, n. 2, Art.6, p. 110 -142, mar.-apr.2015www.bbronline.com.brfunds with

Table 3
, there will be made analyzes of patterns of key financial variables and administrative reported in